Investigating Presence in (VR) Scenario Research: Does presence in a scenario influence willingness to participate in victim offender mediation as a function of crime severity?
J.R. Koetsier S1763814
Faculty of Behavioral, Management and Social Sciences. (BMS) Psychology of Conflict, Risk and Safety
Supervisors: Date:
Dr. S. Zebel 28 Augustus, 2019
Dr. M. Stel
Acknowledgments
The final part of my master degree in Psychology of Conflict, Risk and Safety consisted out of doing research and writing this thesis. During this long journey, I was blessed to meet and work with a lot of very kind and positive people. Especially I would like to thank Dr. S.
Zebel for his supervision, patience and guidance during this project. I would also like to thank Dr. M. Stel for her constructive feedback on this thesis. I have learned a lot on the topic of conducting research. Additionally, I would like to thank the BMS lab, and in particularly Dale Rosen and Peter Slijkhuis, for building the application and providing the equipment I needed for this research project.
Finally, I would like to thank my partner, family and friends for their support and love.
Apeldoorn, Augustus 2019.
Jeroen Koetsier
Abstract
Victim offender mediation is based on restorative justice. Restorative justice involves both the victim and the offender in the process. Victim offender mediation offers victims and offenders of crime the possibility to communicate with each other through mediated contact.
The first aim of the study was to examine how the severity of the offense moderate the relation between presence within a scenario, fear and anger and the willingness to participate in victim offender mediation. The second aim of the study was to examine if presence felt within a scenario could be increased by using a Virtual Reality scenario compared to the often used written scenario in behavioural studies. A total of 196 participants were randomly assigned to either read a scenario (story) or experience the scenario through a Virtual Reality environment, both scenarios victimized the participant in a robbery. To investigate the effect of crime-severity on the relation between presence, fear and anger and the willingness to participate in victim offender mediation, the study used a 2 (written scenario vs virtual scenario) times 2 (high severity versus low severity) between subjects experimental research design. The study found no moderation effect of crime severity on the relation between presence, fear and anger and the willingness to participate in victim offender. The study does suggest that presence can be increased in scenario-research by using VR technology.
Therefore suggesting that there are possibilities regarding the use of VR technology in – scenario based – victim offender mediation research. These results can be used to further improve scenario-based research, making it easier for researchers to gain access to larger sample sizes when participants are hard to reach or find.
Keywords: Victim offender mediation, willingness to participate in VOM, presence, Virtual reality scenario, robbery
Investigating Presence in (VR) Scenario Research: Does presence in a scenario influence willingness to participate in victim offender mediation as a function of
crime severity?
Victim offender mediation is an implementation of restorative justice. Restorative justice provides a different way of thinking about victimization and punishment in general.
Restorative justice views crime as between, and directed by, individual people and thus actively involves the victim and offender in the process (Umbreit, 1998). The current structured system, called retributive justice, views the state as the primary victim and places the victims and offenders in a more passive role. The goal of restorative justice is to (a) empower those affected by the crime to make key decisions, (b) focus more on healing and making justice more transformative, and (c) reduce recidivism (Zehr, 2014). To achieve these goals, victims need to be more involved in the process and be satisfied with the outcome, offenders should understand how their actions affected the victim(s) and take responsibility for those actions, and both the victim and the offender should gain a sense of ‘closure’ so they can reintegrate back into the community (Zehr, 2014). By providing opportunities for dialogue, negotiation and problem solving between the victim and the offender, restorative justice can lead to a conflict resolution, closure for both parties involved and a greater sense of community safety (Umbreit, 1998).
The ideas of restorative justice are also implemented in the Netherlands. Perspectief Herstelbemiddeling (Perspective Restorative Mediation) is a Dutch organisation that offers both victims and offenders of crime the possibility to communicate with each other through mediated contact (Perspectief Herstelbemiddeling, 2017). This mediated contact, under the supervision of a professional mediator, is based on mutual respect between the victim and the offender and is on a voluntary basis. Perspectief Herstelbemiddeling implies that both the victim and the offender can contribute to their recovery through professionally mediated contact (Perspectief Herstelbemiddeling, 2017). For the victims, the mediated contact aims to
contribute to the emotional processing and coping of the incident. It is argued that victims feel more involved and more empowered with the criminal justice process after having participated in mediated contact (Zehr & Gohar, 2003). The aim of mediation is twofold, first and foremost the victim can inform the offender to what extent the crime impacted his or her life. Secondly, the victim can receive information from the offender to answers questions the victim might have about the crime (Zehr & Gohar, 2003). The victim can also be involved in the development of the restitution plan for the offender, which offers the victim influence in making the offender accountable for the crime (Umbreit, Coates, & Vos, 2004). Through mediated dialogue, the offender learns to what extent his or her actions impacted the life of the victim. The mediated contact helps the offender to take responsibility for his or her actions and to develop a fair and reasonable plan to make amends to the victim (Umbreit et al., 2004).
A growing body of research acknowledges the positive effects of victim offender mediation on levels of satisfaction among those who participate. A vast majority of studies reported satisfaction of victims and offenders about the mediation and its outcomes (Davis, Tichane, & Grayson, 1980; Coates & Gehm, 1985; Perry, Lajeunesse, & Woods, 1987;
Marshall, 1990; Umbreit, 1991, 1993, 1994a, 1994b, 1998, 1999; Umbreit & Coates, 1993;
Warner, 1992; Roberts, 1995; Carr, 1998; Roberts, 1998; Evje & Cushman, 2000). These effects were found across different types of offenders, types of victims, different levels of crime seriousness and different cultures. According to a meta-analysis of Umbeit and
colleagues (2004), which stretched over three decades of research, overall satisfaction scores about mediation and resulted agreement typically lay between 80 to 90 percent.
However, around 40 to 60 percent of individuals that were offered the opportunity to participate in victim offender mediated refused this offer. Umbreit and colleagues (2004) explain that the satisfaction found in the literature might thus be caused by the self-selection
factor that overlays research in victim offender mediation. Meaning that, high levels of satisfaction might have something to do with simply being able to choose among justice options in general. Which indicates that the mediated contact itself not necessarily brings about the measured satisfaction levels (Umbreit et al., 2004). A recent study of Jonas- van Dijk, Zebel, Claessen and Nelen (2019) studied this self-selection bias by comparing reoffending rates of three different offender groups: offenders who participated in Victim offender mediation (VOM); offenders who were willing to participate, but whose counterpart declined VOM; and offenders unwilling to participate in VOM. The study concluded that both the willingness of the individual to take restorative steps and the mediated contact itself led to a lower risk of reoffending. Therefore, the study argues that being willing to
participate, the self-selection factor Umbreit and colleagues (2004) describe, does not explain all of the effect VOM has. It suggests that part of the effect of VOM might be due to
processes during the mediated contact itself. As described by Jonas- van Dijk and colleagues (2019), a limitation of their study is that it remains unclear what these processes are. They suggest that further research should use observational studies to examine what happens during mediation and try to understand what the effect of it is. Shapland and colleagues (2008), did such a study by (post-mediation) interviewing offenders and observing mediation sessions. They found that VOM is most effective when: the offender is actively involved; the offender wants to meet the victim; when the mediations makes the offender understand the consequences of the offense and when the mediations is experienced as useful by the offender. However, this observational study was limited in the number of mediation cases observed. The difficulty to observe high numbers of VOM cases can make it troublesome to truly understand the effects of VOM.
A opportunity to this limitation is using scenario-based research methods (see e.g.:
Kippers, 2015; Cefalo 2015; Van Dijk, 2016). Scenario based research methods are often
used to gain insights in the field without having to find and address actual victims and
offenders. Scenarios, also known as vignettes, are short written descriptions of a hypothetical situation. Scenarios are most commonly used to investigate behavioural decision making and behavioural intentions (Gelder et al., 2018). The scenario-based research method is very versatile, it has low costs and is easy to use, making it a widely used method for studying the decision making processes within social sciences (Collett & Childs, 2011). It enables
researchers to study behaviour outside the scope of other data collecting methods. For
example, when the research interest contains sensitive or unethical behaviours, has high costs or has infrequent occurrence in the real world (Collett & Childs, 2011).
It is common in scenario-based research methods that researchers first ask participants to imagine themselves in a described situation or to identify with certain people in the
scenario described. Secondly, when the participants are immersed in the scenario, several questions are asked to investigate how the participants think that they would behave if they would find themselves in such a situation. The reliance on the ability of participants to imagine themselves in a hypothetical situation is seen as a major limitation of the scenario method (Gelder et al., 2018). In other words, the validity of the data collected through this method is influenced by the ability of the participants to imagine themselves in the
hypothetical situation presented to them (Collett & Childs, 2011; Parkinson & Manstead, 1993). The ability and willingness of an individual participant to take a specific perspective within a scenario becomes an undesired variable that potentially influences the collected data.
Another limitation regarding the scenario method lies within the realism of the scenario described. Intense emotions and visceral responses influence and accompany behaviour in real life. Scenarios might not be able to accurately measure the behavior or behavioural intentions of participants when the scenario is unable to elicit the relevant emotional or visceral reactions in the participants (Ditto, Pizarro, Epstein, Jacobson, &
MacDonald, 2006; Gelder et al., 2018). Gelder and colleagues (2018) argue that written scenarios fail to elicit presence (i.e. the extent to which the participant is capable to take the perspective of a person in a certain scenario) and fail to trigger emotions because they include only a small amount of contextual information when compared to real-world situations.
Written scenarios, that try to describe the context as detailed as possible, can only give limited information regarding non-verbal behaviors of others (e.g: facial expressions and body language). Therefore, these scenarios fail to convey potentially relevant contextual information that provide important cues in determining social responses of the participants (Gelder et al., 2018).
To investigate potential solutions to these limitations, the current study examines the use of immersive visual (video-based) scenarios compared to their traditional written
counterpart(s), within the domain of victim offender mediation. The most prominent advantage of visual scenarios is the inclusion of more contextual information (Christian, Edwards, & Bradley, 2010). According to Christian, Edwards and Bradley (2010), video- based scenarios have a higher ecological validity than written scenarios due to the more accurate display of contextual information, giving the scenario a more realistic and real-world feel to the participant. Therefore, video-based scenarios are also capable to create more identification between the participant and the subject of the scenario, which increases the presence felt in the scenario by the participants (Christian et al., 2010). This study uses 360°- video techniques to make the video-based scenario as immersive as possible, since the presence felt by the participant in a video-based scenario is seen as a prerequisite to increase ecological validity (Christian et al., 2010).
This study investigates whether 360°-video-based scenarios are capable of increasing people’s presence when compared to the traditional written scenarios, within the field of victim offender mediation. The scenarios used in this study regard a hypothetical situation in
which the participant is the victim of a robbery. According to theory, victim offender mediation is influenced by the severity of the crime (Zebel, Schreurs & Ufkes, 2017). The severity of the crime seem to influences the willingness of the victim to participate in victim offender mediation and seems to have influences on the emotional evaluation that victims report after the mediation (e.g.: Daly, 2004; 2006; Umbreit, 2002; Zebel, et al., 2017). To examine the effect of crime severity in this study, the robbery has two different versions: a scenario with high severity and a scenario with low severity. Both the high and low severity scenarios are translated to the traditional written scenarios and the 360°-video-based
scenarios. Since the aim of this study is to examine whether 360° video-based scenarios are a more validated data collecting method compared to the traditional written scenarios, it is important that the written scenarios and the 360°-video-based scenarios are as identical to each other as possible (e.g. the same locations, same amount of bystanders present, same time of the day, etc.).
To investigate whether perspective taking has increased among participants in the video-based compared to the written scenario, participants are asked to what extent s/he felt the relevant emotional or visceral reactions and responses, such as feelings of fear and anger.
This research will answer the following two research questions: To what extent is the feeling of being present increased when a video-based scenario is used compared to a written scenario in the field of victim offender mediation? Secondly: how does the severity of the offense moderate the relation between presence, fear and anger and the willingness to participate in victim offender mediation?
Presence
Present research suggest that presence can be experienced in a variety of day-to-day situations. For example, while watching a movie, playing a video game or reading a story (Hartmann et al., 2016). This indicates that there is no need for advanced technology to create a sense of subjective presence. However, presence is assumed to depend on features that the (virtual) environment offers, such as interaction effects and techniques, and the possibility that the medium has to increase realism (Regenbrecht & Schubert, 2002; Shubert & Crucius, 2002). This increase of presence, by rising the realism of the situation, has important features in behavioural studies. It leads to responses that more closely resemble actual responses in the real world (Van Gelder, Otte & Luciano, 2014). Which, as mentioned, means that the behavioural intentions that are provided by the participants have a higher ecological validity when they experience a greater sense of presence in the situation (Gelder et al., 2018). In this study, the realism is increased by using 360°-video based scenario. To investigate if presence can be increased by increasing the realism of the situation, the followings hypothesis is proposed:
H1: The participants in the visual scenario conditions have a greater presence within the scenario compared to the participants in the written scenario conditions.
The role of presence in eliciting fear and anger
One of the more important aspects of presence are the emotional and physiological reactions experienced by the participants. These reactions indicate whether the body and mind of the participant interpreted the virtual reality as a real-world experience. If so, behaviour and behavioural intentions of participants that respond to the virtual reality presented to them can be measured and validated by linking the responses to the appropriate emotional and physical responses seen in the ‘real world’ (Schuemie, van der Straaten, Krijn,
& van der Mast, 2001).
A between-subject experiment that tried to treat subjects with acrophobia (fear of heights) showed an increase in reported anxiety when subjects were exposed to heights in a virtual reality environment (Rothbaum et al., 1995). The same effect remained with other phobias (Schuemie et al., 2001). Regenbrecht and colleagues (1998) also studied the
correlation between presence and fear of heights. The experiment, which included only non- phobic subjects (N=37), showed a weak correlation between anxiety and presence (r=0.251, p> 0.10). However, the study did find that the amount of presence felt was a significant predictor for fear experienced within the sample. Schuemie and colleagues (2000) did a follow-up explorative study (N=10) with subjects that were diagnosed with acrophobia. This study did find a significant correlation between fear and presence felt, but no correlation between presence and reduction of acrophobia. Both studies used questionnaires to test their hypothesis. Another study that examined the relation between fear and experienced presence indicated that experienced presence in a virtual environment can induce the signs of fear people experience when they need to speak publicly, despite the subjects knowing they are speaking in front of a virtual audience (North, North, & Coble, 1998). These studies seem to indicate that emotions are mediated by the presence felt in the scenario. Therefore, the following hypothesis is proposed, which is divided into two sub-hypotheses:
H2a: The visual scenario conditions elicit more fear and anger than the written scenario conditions.
H2b: This effect is mediated by the amount of presence felt by the participant.
Willingness participation VOM (low vs high severe scenarios)
There has been a debate in the literature about the usefulness of severe crimes in VOM (e.g., Zebel, Schreurs &Ufkes, 2017; Dijk, Zebel, Claessen & Nelen, 2019; Larsen, 2014; Richards, 2009; Waltman-Spreha, 2013). It is often thought that victims of severe
crimes are less likely to participate in mediated contact, especially if the crime is committed recently.
There remains inconclusiveness in the effects between the severity of the crime and the willingness for victims to participate in VOM. Umbreit and colleagues (2004) indicates that when victims of severe and violent crimes participate in VOM, the levels of satisfaction after the mediation are exceptionally high. Therefore addressing that VOM can contribute to the victim and the offender in severe cases. However, Umbreit and colleagues (2002) state that the studies that suggest these effects are ‘suggestive at best’. Thus explaining that there is a great need for further research which includes large samples before definite conclusions can be made. They strongly emphasize that mediated dialogue between victims and offenders in severe cases should be victim initiated and be supervised by an experienced and more
advanced trained mediator. There is a greater chance that unintended negative consequences, such as revictimization of the victim, could occur if the mediation of severe cases is offender initiated. The fear of revictimization causes the victim to be less inclined to initiate, or take part in, mediated dialogue (Umbreit et al, 2002). This increases the difficulty to examine the outcomes of mediated dialogue between victims and offenders of severe crimes, on a large scale, since the mediation is always of voluntary nature.
In contrast, a recent study by Zebel and colleagues (2017) found that the offenses in VOM cases were on average more serious than the offenses in the population, when
comparing the average seriousness of crimes to the seriousness of crimes in VOM cases in the Netherlands. Zebel and colleagues (2017) measured the seriousness of the crime by three aspects, (a) the incarnation duration of the offender, (b) the perceived wrongfulness of the victim and (c) the perceived harmfulness of the victim. Therefore, concluding that victims are not discouraged to participate in VOM when they are victimized by a serious/severe crime.
Zebel and colleagues (2017) also found empirical evidence that the willingness to participate
in VOM increased over time in severe cases and that the willingness decreased over time when the cases were less severe. They state that when a victim of a severe or violent crime is asked to participate within a short time-period, the fear of revictimization and concerns of personal safety could outweigh the perceived benefits of the victim to participate in VOM.
However, after a longer time-period, the victim might feel less threatened and remains with questions about the crime, leading to a higher perceived benefit to participate in VOM. A victim of a less severe or violent crime may have less barriers to initiate VOM shortly after the crime took place, but could view the crime as too distant in time or too trivial if VOM is initiated after a longer time-period. More debate regarding this topic comes from findings in the literature which contradict the ‘suggested effect’ Umbreit (2002) speaks of. Daly (2004;
2006), who also evaluated the effects of VOM with victims of severe crimes, indicated that the evaluations of victims of severe cases were less positively; these victims showed less signs of emotional recovery and had a more negative attitude towards the offender when compared to victims of less severe cases. In addition, Pemberton (2012) argued that experiencing a severe crime increases the desire for vengeance and decreases the desire to speak with the offender through mediated contact, especially when the crime was committed recently.
The above-mentioned literature indicates that the severity of the crime and the time between the crime and the proposed mediation can influence the willingness of a victim to participate in VOM. This current study expects that feeling present in a severe or violent crime will decrease the willingness of the victim to participate in VOM, if the victim is asked to participate directly after experiencing the crime. The effect is expected to be the opposite when the victim feels present in a less severe or violent crime and is asked to participate directly after experiencing the crime; increasing the willingness of the victim to participate.
These expectations are in line with the results found by Zebel and colleagues (2017).
Furthermore, this study hypothesizes that the severity of the crime moderates the positive relation between presence, fear and anger and the willingness of the victim to participate. The moderating effect is expected to either be positive or negative depending on whether the crime experienced by the victim was of high or low severity. To study this effect, the following hypothesis is proposed, which is divided into two sub-hypotheses:
H3a: There is a positive relation between presence-, fear- and anger felt within the scenario and the willingness to participate in victim offender mediation.
H3b: The severity of the offense moderates the positive effect of presence-, fear- and anger felt within the scenario on the willingness to participate; the effect is expected to be less positive (or negative) when the offense is of high severity.
Conceptual model
Figure 1 shows the conceptual model in which the hypotheses of this study are reflected.
Figure 1: Conceptual Model
Method Participants
In total 196 people (95 men, 89 women, 12 unknown) participated in the study. The
of the participates were: Dutch (N=129), German (N=31) and Other (N=24) (missing N=12) . The participants needed to be at least eighteen years old and had to agree with the informed consent form, which was the only selection criterium to enter the study. Not all the
questionnaires were filled in completely. Therefore, not all participants are included in the entirety of the research. Only participants that filled in the data for the specific subject are taking into account when calculating the data and conducting the analyses. Therefore, some analysis have a higher missing N then others.
The participants in the control, written scenario condition, were recruited using
convenience - and snowball sampling. Firstly, the participants who were in close proximity to the researcher were recruited. Secondly, these participants were asked if they knew more people who were willing to participate that met the age criteria of the study. Thirdly, by sharing an anonymous link on social media participants could participate in the study via that link.
Since the experimental, visual scenario condition, could not be conducted via online- survey, the researcher needed to be physically present in the same room with the participants in the experiment to conduct the research. Therefore, the participants in the experimental, visual scenario condition, were recruited by asking random bystanders (face-2-face) if he/she would be willing to participate in the study. The researcher asked randomly selected students of the University of Twente. If the participant was willing to participate in the study, the researcher and participant went to an enclosed room where the visual scenario condition of the research was conducted.
Overview and Design Research design
The study has a 2 (written scenario vs visual scenario) times 2 (high severity versus
low severity) between subjects experimental research design. Figure 2 illustrates the design of this study.
Figure 2: Research Design
All scenario conditions displayed the participant as a victim of a robbery. The participants in the written-scenario condition could open their scenario via a link and fill in their questionnaires online. Making it possible for these participants to participate from their home. The participants that were asked by the researcher, and agreed to participate to the video-scenario, all went into the same rented enclosed room to conduct the visual scenario.
The participants were given an informed consent form and were informed about the video- technique used in the study before participating in the visual-scenario condition. After the video-scenario was displayed, the participants immediately filled out the questionnaire.
The questionnaire used for the visual scenario conditions differ from the written scenario condition. The visual scenario conditions asked some questions to measure whether the visual-scenario was conducted properly (did the participant see the offender, did the
also measured the variable ‘simulation sickness’. Simulation sickness can only occur when a person is in a VR-environment, this variables was therefore excluded in the written scenario conditions of the study.
The four conditions (high vs low severity & written vs visual scenario) are similar in the event and the crime itself. The victim is robbed while withdrawing cash from an ATM machine. The differences between the scenarios are the following: (A) the toon of voice of the offender: in the high severity (HS) scenario the robber uses threatening (body) languages and threatens to kill the victim if s/he does not give him the money, in the low severity (LS) scenario the robber uses no languages. (B) violence used: in the HS scenario the robber punches and grabs the victim and hits the victim with his gun, in the LS scenario the robber only pushes the victim to the ground so he can grasp for the money. (C) And by making the robbery seem either planned (high severity), by having the offender stare and wait for the victim, or more like a ‘drift’ (low severity) where the offender walks straight up towards the victim.
Procedure and Manipulation Written-scenario condition
The participants were sent a link in which they could open the scenario and the
questionnaire. The scenario was manipulated in the severity of the robbery, meaning that they either received a high-severe or a low-severe scenario. In the high-severe condition, the emphasize of the text directed the participants attention more towards the aggression of the offender, his threating tone, wording, the presence of a gun and body languages in
comparison to the low-severe condition. In the low-severe condition there was no verbal contact with the offender, the offender did not use as much aggression and did not had a gun.
The scenarios were directly followed by the questionnaire. The written scenarios can be found in attachment 1.
Video-scenario condition
Every participant was randomly assigned to either the low-severe or high-severe video-scenario. Once the participant agreed to participate in the study, an informed consent form was given. In which the participants were also informed that the virtual environment in which they were going to be placed, displayed a hypothetical robbery. After the participants signed the informed consent, a brief instruction about the 360°-video technique used in the study were given. The participant can ask questions about the video techniques used to the researcher if they have any. Then s/he is placed in the virtual environment in which the scenario is displayed (see figure 3 and 4 for examples on how the offender was portrayed in each severity condition). Directly after the scenario is finished, the participant is asked to fill out the questionnaire.
Figure 3: Screenshot High Severity Condition
Figure 4: Screenshot Low Severity Condition
Questionnaire
A different questionnaire was used for the different scenario-conditions in the study (written vs video) (see 2.2 research design). All questionnaires measured the presence felt in the scenario, the fear and anger of the participants and how severe the participant experienced the scenario. After these variables were measured, the participants were given a brief
instruction which informed them about Victim Offender Mediation and its options. After which the questionnaire continued and asked the participants to what degree they would be willing to participate in VOM themselves, if they were victimized and would be given the option. The following sections explains in depth how the questionnaire was used to measure the variables.
Dependent Measures
The following variables were measured in this study: ‘Presence felt in the scenario’,
‘Fear’ and ‘Anger’, ‘Severity’ and ‘The Willingness to Participate in VOM’. First, this
section explains how these variables are measured in the questionnaire used in the study, Secondly, the researcher tested if the different severity conditions were manipulated in the desired way: the severity is higher in the high severity conditions and lower in the low severity conditions. Thirdly, the research design of the study is explained.
Measuring presence
The most commonly used method to measure presence are questionnaires. These questionnaires measure the subjective rating of presence felt by the subjects in the scenario.
Since the subjective nature of these observations can be biased in more than one way, the measured responses could be less reliable (Schuemie et al., 2001). Therefore, to make the measurements more reliable, multiple questions are used in this study to measure the same construct.
An often-used questionnaire to measure presence in a VR environment is the Igroup Presence Questionnaire (IPQ) (Schubert, Friedmann and Regenbrecht 2001). This
questionnaire, consisting of 14 items, is based on three aspects of presence (a) Spatial Presence; the sense of being physically present in the VR environment, (b) Involvement;
measuring the attention devoted to the VR environment, (c) Experienced Realism; measuring the subjective experience of realism in the VR environment. The IPQ is a well-established questionnaire which has been tested and improved throughout the years to measure presence in research and has seen use by pioneers in the field (alfa = .85) (e.g.: Gelder et al., (2018)).
To measure presence in this study, a rephrased and altered version of the IPQ was used. The questionnaire was rephrased to fit all conditions in the study. This was necessary because the IPQ is originally used to measure presence only in a virtual environment (read visual
scenario), not within a written-scenario. To give an example of such a rephrased question;
The IPQ has a question formulated the following: ‘How aware were you of the real world surrounding while navigating in the virtual world? (i.e. sounds, room temperature, other
people, etc.)?’ Since this questions does not fit a written scenario, the question was altered to this: ‘While experiencing the scenario, I was less conscious of my true surroundings
(example: sounds, room temperature)’
To add more theoretical body to the questionnaire, two items of the questionnaire developed by Usoh and colleagues (2000) were also rephrased and used. This questionnaire is based on multiple questions which all are variations of three major themes: (a) the subjective sense experienced by the subject of ‘being there’, (b) the extent to which the virtual
environment becomes more real or present than everyday reality, and (c) the ‘locality’, which measures to what extent the virtual environment is thought of as a ‘place’ that was
‘visited’ rather than just a set of images that were seen, or a video that was watched (Usoh et al., 2000). These aspects have overlapping elements with the IPQ, both are focused on the subjective experience the subject has while being in the virtual environment.
The same components of the IPQ were found in this study, presence was divided in three different scales based on a factor analysis: 1: Presence SP (Spatial Presence, 6 items) (α= .805), which indicates to what extent the participant had the feeling to be present in another location. 2: Presence Realism (Experienced Realism, 3 items) (α= .797), this score indicated to what extent the participant found the scenario realistic and 3: Presence Inv (Involvement, 2 items) (R= .288**), which measured the involvement of the participant in the scenario. The results of this factor analysis can be found in table 1.0.
Measuring fear and anger experienced in the scenario
The fear and anger of the participant is measured in the questionnaire by using 5 items. The items asks the participant how they would feel after the robbery. All items are asked in a 1 to 7 Likert scale format (1 = strongly disagree, 7= strongly agree), asking the participant to what extent they agree with feeling the following emotion after the robbery: ‘As
a victim in this scenario, after the robbery I would feel…..’ ‘outraged’, ‘panic’, ‘frightened’,
‘anger’ or ‘fear’.
A factor analysis showed that ‘panic’, ‘frightened’ and ‘fear’ measured the same construct;
fear (α= .799). ‘Anger’ and ‘outraged’ measured, according to factor analysis, the construct;
anger (R= .467**). The results of this factor analysis can be found in table 2.0.
Measuring the perceived severity of the crime
It is important to measure whether severity is adequately manipulated between the scenarios in the study. Therefore, 8 questions are asked that control this manipulation. Zebel and colleagues (2017) conceptualized the severity of crimes by three constructs; the
‘perceived wrongfulness’, the ‘perceived harmfulness’ and the ‘incarnation time’ of the offender. Since this study uses hypothetical crimes, the incarceration time of the offender is not included in the manipulation. However, the perceived harmfulness and perceived wrongfulness are taking into account to perform a manipulation check of the severity of the crime between the different conditions. Zebel and colleagues (2017) indicate that the perceived wrongfulness is the normative evaluation that the victim has of the offence, the perceived harmfulness is seen as the degree to which the victim feels that s/he is (materially or immaterially) harmed by the offence. All items are measured in a 1 to 7 Likert scale (1=
strongly disagree and 7= strongly agree).
To illustrate how perceived harmfulness and perceived wrongfulness are measured in this study, some example questions are shown below:
Perceived wrongfulness: ‘As the victim in this scenario, after the robbery I would think... - What the robber did was morally wrong’ and ‘As the victim in this scenario, after the robbery I would think... - The robber deliberately harmed me’.
Perceived harmfulness: ‘As the victim in this scenario, after the robbery I would feel... - Emotionally damaged by the offense’ and ‘As the victim in this scenario, after the robbery I would feel... - Physically damaged by the offense’.
A factor analyse of perceived wrongfulness indicated that only 1 construct was measured (3 items in the scale, α= .419). A factor analyse of perceived harmfulness also indicated that 1 construct was measured (3 items in the scale, α = .537). The item: ‘as the victim in this scenario, after the robbery I would feel … - financially damaged by the offense’
measured a different construct and was therefore removed from the perceived harmfulness scale.
The last item that measured severity in this study examined the participants own evaluation of the severity level of the offense: ‘please score how severe you find the offense in general - The general severity of the offence was...’ on a 1 to 7 Likert scale (1= not severe at all and 7= Extremely severe).
Measuring the willingness to participate in VOM
The willingness to participate in VOM is measured using 13 items in the
questionnaire. All items are measured on a 7-point Likert scale (1= strongly disagree and 7=
strongly agree). The participants are informed about VOM and its implications for victims prior to answering these items. This information gives participants a general idea of VOM and its implementation for victims (see attachment 3). The information and the questionnaire are distributed to the participants directly after s/he experienced the scenario, making the time-period between the ‘offence’ and the evaluation of the participant whether s/he would be willing to participate in VOM as brief as possible.
The items measure the attitude of participant towards VOM in general (example:
‘Please indicate to what extent you agree with the following statements about Victim
Offender Mediation (VOM) - I think that VOM is pointless’) and how likely the participant would participate in VOM, if the robbery that s/he hypothetically experienced in the scenario would happen to them in real life (example: ‘When asked if I would be willing to participate in Victim Offender Mediation.... - I would not be willing to participate’).
The items regarding victim offender mediation where divided into three separate scales using reliability analyses and factor analyses: VOM participation (6 items , α= .876) VOM doubt (2 items, R= .468**) and VOM reason (2 items, R= .392**) The results of this factor analysis can be found in table 3.0. In total, 3 items were removed from analyses. These items were removed because they either lowered the reliability of the analyses and/or did not measure an intended construct. The first two items both measured to what extent the
participant would not be unwilling to participate because they are either to fearful or to angry to see the offender: ‘I would not be able to control myself, I am afraid I would attack the offender’ and ‘I think I would be too fearful of the offender to participate in VOM’. These items were removed since they decreased the reliability of the analyses. The other item that was removed measured if the participant only wanted to participate in VOM if the contact is face-to-face: ‘I would only be willing to participate, if the contact is face-to-face’. This item seems to stand alone in the analyses since it is the only item that measures if there would be a preference in communication channels. Since this item has no connection to other items in the scale, the item is removed.
Results Descriptive statistics and correlations
Means, standard deviations, reliabilities and intercorrelations for all used variables are shown in Table 4.0.
Table 4.0. Means, standard deviations, reliabilities and intercorrelations among variables
Variables M SD α 1 2 3 4 5 6 7 8 9 10 1
1. 1. Presence SP 2. 2. Presence Realism 3. 3. Presence Inv 4. 4. Fear
5. 5. Anger
6. 6. Perceived wrongfulness 7. 7. Perceived harmfulness 8. 8. VOM participation 9. 9. VOM reason
10. VOM doubt
4.49 5.00 4.60 4.59 5.31 5.46 4.88 4.86 5.03 3.50
1.04 1.13 1.12 1.34 1.15 .95 1.03 1.32 1.33 1.47
.805 .797 - .799 - .419 .537 .876 - -
-- 512**
.571**
.161* .114 .145* -.017 .054 .021 -.083
--- .385**
.221**
.210**
.206**
.095 .007 .004 -.066
--- .155* .056 .122 .076 .143 .078 -.081
--- .207**
.344**
.423**
.058 .288**
-.273**
--- .221**
.109 .157* .073 -.056
--- .280**
.003 .126 -.075
--- -.015 .221**
-.162* --- .697**
.047 ---
-.197** ---
Note. *p <.05, ** p <.01, α = before standardizing,
Table 4.0 shows some interesting statistics. It shows that the three variables to measure presence are significantly positively correlated with each other. Thus indicating that the realism of the scenario, the level of involvement the participate feels in the scenario and the feeling of being physically present in the scenario have a significant relationship. Fear elicit by the scenario was positively correlated with all of the presence scales. Indicating that the presence felt in the scenario has a relationship with eliciting fear. Anger was correlated with realism in the scenario, but was not associated with the spatial, feeling physically present in the scenario, and involvement in the scenario. Anger was significantly positively correlated with fear. Indicating that the scenarios did not elicit one singular emotional response, but influenced both emotions in the same direction. Presence SP, presence realism, fear and anger were significantly positively correlated with the perceived wrongfulness. Which suggest that feeling presence in the scenario and eliciting the emotional responses has a relationship with morally reprehensibility of the scenario. The perceived harmfulness, the degree to which the victim feels that s/he is (materially or immaterially) harmed by the offence, is positively correlated with fear and the perceived wrongfulness. The willingness to participate in victim offender mediation is positively correlated with anger. This partly rejects hypothesis 3a, since this indicates that there is no relation between presence and fear and the willingness to
participate in victim offender mediation. The reasoning to participate in victim offender mediation is positively corelated with fear, perceived harmfulness and willingness to
participate. Doubt to participate in victim offender mediation is negatively correlated to fear, perceived harmfulness and to reasoning to participate in victim offender mediation.
Randomisation check
To examine whether there was a difference between the groups in education completed, a chi-square analysis was conducted. The participants in the written scenario
condition had a significantly higher completed education (m=3.25, SD= .11) than the participants in the VR condition (m=2.81, SD= .12 ). There were no educational differences between the different severity conditions. To calculate if there was also a significant
difference in the age between the groups, a ANOVA was conducted. The participants in the written scenario condition were significantly older (m = 31.83, SD= 1.07) than the
participants in the VR condition (m =21.50, SD= 1.17) F(1,182) = 42.74, p = <.001. There were no age differences between the different severity conditions. Because of these
differences, age and educational levels are controlled for when conducting the analyses.
Manipulation of severity
In the study severity was manipulated so that participants in the high severity conditions were to expect to report a higher severity score than participants in the low severity conditions. To investigate if the manipulation of severity was successful, a manipulation check was conducted using 3 ANCOVA’s, measuring the effects of the manipulations of severity and scenario on the dependent variables; perceived harmfulness, perceived wrongfulness and the general severity score.
Perceived harmfulness
There is a significant effect of severity on perceived harmfulness; b= 45.802, F(1,173)=33.03, p= <.001. ηp2= .16.. Low severity mean = 4.04, SD= 1.29, high severity mean = 5.03, SD= 1.15. Indicating that participants in the high severity group had
significantly higher perceived harmfulness scores compared to the low severity group. The model shows that scenario also had a significant effect on the perceived harmfulness scores;
b=10.29, F(1,173)= 7.42, p= .007. ηp2= .041. The written scenario conditions have an overall higher mean then the VR conditions: written scenario m= 4.84, SD= 1.33; VR scenario m=
4.22, SD= 1.21. Which indicates that there is a significant higher perceived harmfulness score
in the written scenario condition, compared to the VR condition. There is no interaction effect between scenario and severity. None of the covariables had a significant effect on the
perceived harmfulness.
Perceived wrongfulness
Severity had a moderate effect (p=<.010) on perceived wrongfulness; b= 3.01, F(1,173)= 3.392, p= .067. The low severity mean was higher than the high severity mean:
low severity m= 5.97, SD= .93; high severity m= 5.72, SD= 1.04. Indicating that the intended manipulated effect, higher severity group has a higher perceived wrongfulness scores, is not supported in this test. Scenario had a significant effect on perceived wrongfulness; b= 17.878, F(1,173)= 20.166, p= <.001. The written scenario had a mean score of 5.62 on perceived wrongfulness and a SD of 1.13 and the VR scenario had a mean score of 6.11 on perceived wrongfulness and a SD of .73. Indicating that the VR scenario conditions had a significant higher perceived wrongfulness scores that the written scenario conditions. There is no
interaction effect between scenario and severity. None of the covariables had a significant (p=
<.005) effect on perceived wrongfulness.
General severity
No significant differences were found on the general severity score between the groups and the covariables.
This indicates that the study only partially succeeded in manipulating the severity as
intended. The perceived harmfulness was properly manipulated. The perceived wrongfulness gave a unexpected reversed effect. Although the effect is not significant on a 95% confidence level (p=.067), the test shows that there is a lower score in the high severity group compared
to the low severity group. The manipulation of severity, and its weaknesses will be discussed in more detail in the discussion of this study.
Effects of scenario on presence
The researcher expected that the participants in the VR condition would report a higher presence felt compared to participants in the written condition (hypothesis 1). To examine if the expectation of the researcher came true, an ANCOVA was conducted to compare the presence between the written scenario and the VR scenario.
Presence Spatial
Presence SP was reported higher in the VR condition (m = 4.76, SD= .93) then in the written condition (m = 4.31, SD= 1.07). This differences was significant b= 8.649, F(1, 173)= 8.631, p= .004. There was no interaction effect between scenario and severity. None of the other covariables had a significant effects on presence SP. This indicates that, as
expected, the participants felt a higher spatial presence in the VR condition compared to the written condition.
Presence Realism
Unexpectedly, no significant differences were found on the presence realism scores between the groups and the covariables.
Presence Involvement
Presence Inv scored significantly higher in the VR condition (m= 4.85, SD= 1.14) then in the written condition (m= 4.43, SD= 1.14); b= 5.549, F(1,173)= 4.302, p= .040. This ANCOVA indicates that the participants reported more involvement in the scenario in the VR conditions compared to the written condition. There was no interaction effect between
scenario and severity. None of the other (co)variables had a significant effects on presence Inv.
These results are partly in agreement with hypothesis 1, giving indication that the VR
condition leads to a higher sense of spatial presence and more involvement of the participant in the scenario. However, it does not indicate that the realism of the scenario increases, compared to the written counterpart. The means and standard deviations are illustrated in Table 5.0
Presence mediating the effect of scenario on fear and anger
According to classic methods paper of Baron & Kenny (1986), mediation is only possible when there is a significant total effect between X and Y (c path). The total effect of scenario on fear and the total effect of scenario on anger was calculated using the Process addon by Hayes, module 4, while controlling for the differences in age and education between the groups. Both tests indicated that there is no total effect of scenario on fear or anger, while controlling for the differences in age and education: scenario on fear; b= -.1989, t(6,175) = -.8704, p = .385, scenario on anger b= -.2410, t(6,175) = -1.2227, p = .223. Since presence cannot mediate the effect of scenario on fear or anger, hypothesis 2 is therefore rejected.
Analysing the willingness to participate in Victim Offender Mediation
When the participants are asked if they would be willing to participate in VOM, 72.8% indicate to somewhat agree, agree or strongly agree with participation, 15.8%
somewhat disagreed, disagree or strongly disagree with participating and 11.4% indicated to be neutral.
An analysis of covariance (ANCOVA) was conducted to examine the effects of scenario and severity on the willingness of participants to participate in VOM, while controlling for differences in age and education between the groups.
VOM Participation
There were no significant effects of the manipulations : severity b= .523, F(1,173) = .296, p= .587, scenario b= .519, F(1,173) = .294, p= .588. There was also no interaction effect: b=1.103, F(1,173) = .625, p= .430. Which indicates that a high severity does not decreases the willingness to participate in VOM. Age did not have an effect on the willingness to participate in VOM: age b= .003, F(1,173) = .077, p=.781. None of the education levels scored a significant difference at the 95% significant level. However, participants that indicated to have a PhD or higher form of education had a marginally different score (p=<.010): b= -1.742, F(1,173) = 2.989, p= .086. Indicating that participants with a PhD or higher completed education scored marginally lower on the willingness to participate then the control dummy variable, people who indicated that high school was there highest completed educational level.
To examine if other factors contribute to the decision-making of the participants to be willing to participate in victim offender mediation, VOM doubt and VOM reason were also analysed as dependent measures. Two analysis of covariances were conducted to see if these variables were influenced by the difference in scenario conditions and/or the differences in the severity, when controlling for the differences in age and education.
VOM Reason
The different scenario and/or severity conditions did not influence the reasoning of the participant to participate in VOM: scenario b= .091, F(1,173) = .054, p= .817; severity b=
1.366, F(1.173) = .808, p= .370. There was no interaction effect: b= 1.915, F(1.173) = 1.133, p= .289. Age did not seem to effect the reasoning: b= .008, F(1.173)= .533, p= .466.
However, the test indicates that participants that have a PhD or higher level of education does score significantly lower than participants with only a high school completed education: b= - 3.243, F(1.173)= 18.272, p = .001. This indicates that participants with a PhD or higher level
of education seem to find less reason to participate in VOM than participants that completed only a high school education.
VOM Doubt
The scenario conditions did not have a significant effect on VOM doubt, scenario b=
3.658, F(1,173) = 1.797, p=.182; Severity also did not influence the amount of doubt the participant had to participate b= 5.110, F(1,173) = 2.511, p=.115. There was no interaction effect: b= .048 F(1,173) = .024, p= .878. Educational level did not influence the amount of doubt participants had. The age of the participant did have a significant positive effect on the amount of doubt a participant has when asked if they are willing to participate in VOM: b=
.028, F(1,173) = 5.893, p= .016. Which indicates that the doubtfulness of the participant experienced in the study was effected by age.
The moderation effect of severity on the relationship between presence and victim offender mediation.
The moderation effect of severity was calculated using the PROCESS addon by Hayes, module 1. To calculate the moderation effect of severity on the relationship between presence and victim offender mediation, a total of 9 analyses were conducted. All three scales to measure presence (presence SP; presence Realism; presence Involvement) were used as independent variables in the analyses. The three scales measuring willingness to participate in victim offender mediation (VOM participation; VOM reason; VOM doubt) were used as dependent variables in the analyses. The interesting results found in the data will be discussed in this section.
VOM participation
There was a main effect of presence involvement on VOM participation b=.176, t(173)= 2.028, p= .044. This indicates that an increase of presence involvement increases
willingness of the participant to participate in victim offender mediation. Severity did not have an effect on VOM participation b= -.156, t(174)= -.793 p= .429. Severity also does not seem to moderate the positive significant relationship between presence inv and VOM participation b= -8.02, t(173)= -.131, p= .896.
VOM reason
The data indicates that the predictors: presence realism, severity and the covariables:
educational level and age, had a moderate effect on VOM reason F(8,173)= 1.852, p= .071, R2= .079. There is no main effect of presence realism on VOM reason: b= .022, t(173)= .253, p= .800. Furthermore, the data suggest that there is no effect of severity on VOM reason: b=
.156, t(173)= .810, p= .419. However, there seems to be a moderate moderation effect: b= - .3088, t(173)= -1.81, p= .073. Addition of the interaction was a moderate change to the model: F(1, 173)= 3.26, p= .073, R2 change =.017. See figure 5.
Figure 5: Moderation Effect of Severity on the Relationship between Realism and Reasoning to
participate
The graph illustrates that when presence realism increases in the low severity group, VOM reason increases. The opposite is happening in the high severity group, when presence realism increases in the high severity group, VOM reason scores decrease. This means that in the higher severity group, participants score less positive to reasonings to participate in VOM as the perceived realism of the scenario increases, which is in line with the expectations of the study.
Educational levels does seem to have an significant impact in this analysis. There is a significant lower score of VOM reason in the PhD educated group, compared to the
reverence group, high school completed education: b= -3.19, t(173)= -3.26, p= .001. All other education levels did not have an effect on VOM reason scores.
VOM doubt
The overall model indicates that VOM doubt does seem to be influenced by the predictors, presence SP, severity and the covariables: educational level and age, in the model, F(8, 173)= 2.245, p= .027, R2 = .307. However, presence SP and severity are not a significant predictor in the model, presence SP: b= -.090, t(173)= -.847, p= .399; Severity: b= -.347, t(173)= -1.63, p= .105. There is also no interaction effect, b= -.227, t(173)= -1.069, p= .287.
Age did have a moderate effect on VOM Doubt: b= .020, t(173)= 1.825, p= .070. Which indicates that for every 1 unite increase in age, the VOM doubt score increased by 0.201.
Giving an indication that older participants were more doubtful in participating in victim offender mediation than younger participants.
When analysing what effect severity and presence realism have on VOM doubt, the overall model of process indicates that there is a significant effect, if the covariables:
educational level and age, are taking into account : F(8, 173)= 2.359, p= .020, R2 = .0983.
Again, there seems to be no main effect of presence realism and of severity on VOM doubt:
Presence realism, b= -.1200, t(173)= -1.260, p= .2093; severity, b= -.336, t(173)= -1.585, p=
.115. And no interaction effect: b= -.173, t(173)= -.920, p= .359. And again, age seems to have an significant effect on the VOM doubt score: b=.0226, t(173)= 2.083, p= .039.
The data gives the same impression when examining the effect of severity and presence involvement on VOM doubt, again there seems to be an overall effect of the model of process if all covariables are taken into account, F(8, 173)= 2.030, p= .046, R2 = .0858.
Again there is no main effect of presence involvement or severity on VOM doubt: Presence involvement, b= -.0329, t(173)= -.347, p= .729; severity, b= -.3317 t(173)= -1.549, p= .123.
And no interaction effect: b= -..0609, t(173)= .3162, p= .752. And again, age seems to have an significant effect on the VOM doubt score: b=.0229, t(173)= 2.065, p= .040.
These analyses indicate that VOM doubt is mostly effected by the age of the participant, where the older participants are more doubtful to participate in victim offender mediation.
These results are in agreement with hypothesis 3. A main effect was found of presence involvement on participation in victim offender mediation. Meaning that feeling involved in the scenario, increased the willingness of the participant to participate in victim offender mediation, which is in line with hypothesis 3a. There was also a moderate
moderation effect found of severity on the relationship between realism and reasoning to participate, which is in agreement with hypothesis 3b; that severity moderates the effect presence has on willingness to participate. The effect was, as expected, that the high severity decreased the positive relation between presence involvement and reasoning to participate in victim offender mediation. However, the moderation effect was not significant on a 95%
confidence level (p= .073) and the effect was only found on the reasoning for participants to participate, not the actual willingness of the participant to participate.
The moderation effect of severity on the relationship between fear and victim offender mediation.
To calculate the moderation effect of severity on the relationship between fear and victim offender mediation, a total of 3 analyses were conducted. Fear was always the
independent variable. The three scales measuring willingness to participate in victim offender mediation (VOM participation; VOM reason; VOM doubt) were used as dependent variables in the analyses. The interesting results found in the data will be discussed in this section.
VOM reason
When all predictors are taking into account in the model, the overall model of the moderation effect of severity on the relationship of fear and VOM reason shows a significant effect: F(8,173)= 2.952, p= .004, R2 =.1201. The main effect of fear on VOM reason is also significant, b=.2483, t(183)= 3.392, p= <.001. Indicating that more fear leads to more reasoning to participate in victim offender mediation. There is no significant main effect of severity on VOM reason and there was also no significant interaction effect. Education did have a significant effect on the VOM reason score, participants with a PhD or higher
completed education scored significantly lower on the scale compared to the participants that completed high school as there highest form of education, b= -2.455, t(173)= -2.500, p= .013.
Age did not have a significant effect on the VOM reason scores.
VOM doubt
There is a significant effect found in the overall model, including all predictors, of the moderation effect of severity on the relationship of fear and VOM doubt: F(8,173)= 3.334, p= .001, R2 = .1336. Fear has a main effect on VOM doubt, b= -.251, t(173)= -3.10, p= .002.
Indicating that for every 1 unite increase in fear there is a -0.251 decrease in VOM doubt score. There is no significant main effect of severity on VOM doubt and there was also no significant interaction effect. Age remained a significant predictor of VOM doubt scores, indicating that a higher age leads to more doubt in participating in this study, b= .021, t(173)=
1.993, p= .048.
These results are partly in agreement with hypothesis 3. A main effect was found of fear on VOM reason and on VOM doubt. Meaning that more fear leads to more positive view on reasoning to participate in victim offender mediation and that more fear leads to less doubt in participating in VOM mediation, which is in agreement with hypothesis 3a. However, no moderation effect was found of severity on the relationship between fear and victim offender mediation. Which rejects hypothesis 3b.
The moderation effect of severity on the relationship between anger and victim offender mediation.
To calculate the moderation effect of severity on the relationship between anger and victim offender mediation, a total of 3 analyses were conducted. The three scales measuring willingness to participate in victim offender mediation (VOM participation; VOM reason;
VOM doubt) were used as dependent variables in the analyses, anger was used as the independent variable. The interesting results found in the data will be discussed in this section.
VOM participation
The main effect of anger on VOM participation is significant, b= .1817, t(173)=
2.041, p= .043. This indicates that in this study more anger leads to a higher participation in victim offender mediation. There is no significant main effect of severity on VOM
participation and there was also no significant interaction effect. Again, educational level does seem to influence participation in victim offender mediation. The participants that had completed a PhD of higher education scored significantly lower on participation than the participants that completed only a high school educational level, b=-1.982, t(173)= -1.991, p=
.048.
VOM reason
The overall model of the moderation effect of severity on the relationship of anger and VOM reason shows a moderate effect: F(8,173)= 1.91, p= .061. There is no main effect of anger or severity on VOM reason. There is also no interaction effect. As seen before, educational levels did have a significant effect. The direction of the effect is the same as described before; the participant with the highest completed educational level scored
significantly lower than the participants that completed only a high school educational level on VOM reason, b= -3.397, t(173)= -3.449, p= <.001.
VOM doubt
The overall model of the moderation effect of severity on the relationship of anger and VOM doubt shows a significant effect when all predictors are taken into account: F(8,173)=
2.543, p= .012, R2 = .105. This indicates that 10,52% of variance in the VOM doubt scores are explained by the IV’s; anger, severity and the covariables; age and the dummy coded educational levels. There is a moderate main effect of anger on VOM doubt: b= -.1870, t(173)= -1.933, p= .055. Meaning that an increase in anger leads to decrease in VOM doubt.
There is also a moderate effect of severity on VOM doubt: b= -.4204, t(173)=-1.954, p=.052.
Meaning that a higher severity leads to a decrease in VOM doubt scores. There is no
interaction effect found in the analysis. The covariables age and education both had effect on the VOM doubt scores, age had a positive significant effect b= .0233, t(173)= 2.159, p= .032.
Indicating that participants with a higher age scored significantly higher on the VOM doubt scale. The participant with the highest completed educational level, PhD or higher, scored moderately higher on VOM Doubt, then the participants that completed a high school education as their highest education b= 1.960, t(173)= 1.812, p= .071.
These results are partly in agreement with hypothesis 3. A main effect was found of anger on VOM participation and a moderate main effect was found of anger on VOM doubt.
This indicates that in this study more anger leads to a higher participation in victim offender
